arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2605.05751cs.DC
收藏

面向边缘智能的隐私保护机器学习框架:实证分析

A Privacy-Preserving Machine Learning Framework for Edge Intelligence: An Empirical Analysis

Western Sydney University(西悉尼大学)

Quoc Lap Trieu, Bahman Javadi, Jim Basilakis

AI总结:

本文提出一种面向边缘智能的隐私保护机器学习框架,评估了差分隐私、安全多方计算和全同态加密在模型精度、响应时间和能耗方面的性能,揭示了隐私保护与系统效率之间的权衡。

摘要:

随着边缘智能(EI)在智能医疗、制造业和关键基础设施中的普及,确保数据隐私的同时维持系统效率成为日益严峻的挑战。本文提出一种新的隐私保护机器学习(PPML)框架,专门针对EI应用,包括四层系统架构和训练与推理算法。我们重点研究了三种主流方法:差分隐私(DP)、安全多方计算(SMC)和全同态加密(FHE),并评估它们对关键性能指标的影响,包括模型精度、响应时间和能耗。实现实验和广泛的跟踪基于模拟结果表明,DP通常在吞吐量和延迟方面接近明文基线,但随着模型复杂度增加,精度下降(在AlexNet上可达35%,在LeNet上低于18%的FordA)。SMC性能由通信驱动;网络带宽和轮次复杂度决定了端到端延迟。对于AlexNet,将链路容量从250 Mbps增加到500 Mbps可将延迟减少约30%。FHE对模型结构和数值精度位宽高度敏感,更紧的参数导致计算开销显著增加;我们观察到响应时间比DP高出约1000倍。除了效率外,DP通过减少攻击者在黑盒模型窃取中的数据效率,推动了隐私-效用-可提取性前沿。而SMC和FHE在推理过程中保护输入和参数,但需要互补的输出控制以达到相似的抗提取性。这些发现为边缘计算场景中隐私、性能和资源效率之间的权衡提供了关键见解。

英文摘要:

As Edge Intelligence (EI) becomes increasingly prevalent in domains such as smart healthcare, manufacturing, and critical infrastructure, ensuring data privacy while maintaining system efficiency is a growing challenge. This paper presents a new privacy-preserving machine learning (PPML) framework tailored for EI applications, including a four-layer system architecture and training and inference algorithms. We focus on three leading approaches: Differential Privacy (DP), Secure Multi-party Computation (SMC), and Fully Homomorphic Encryption (FHE), and assess their impact on key performance metrics, including model accuracy, response time, and energy consumption. Results from real implementation and extensive trace-based simulations of inference tasks show that DP generally preserves throughput and latency close to plaintext baselines, while accuracy drops with model complexity (up to 35 percent on AlexNet and under 18 percent on LeNet for FordA). SMC performance is driven by communication; network bandwidth and round complexity determine end-to-end latency. For AlexNet, increasing link capacity from 250 Mbps to 500 Mbps reduces latency by about 30 percent. FHE is highly sensitive to model structure and numerical precision bit width, with tighter parameters imposing substantial compute overhead; we observe roughly a 1000 times increase in response time compared to DP. Beyond efficiency, DP shifts the privacy-utility-extractability frontier by reducing the attacker's data efficiency in black-box model stealing, whereas SMC and FHE, while protecting inputs and parameters during inference, require complementary output controls to achieve similar resistance to extraction. These findings provide critical insights into the trade-offs between privacy, performance, and resource efficiency in edge computing scenarios.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑